ML LFAST: Difference between revisions

From VASP Wiki
No edit summary
No edit summary
Line 9: Line 9:
One can still have Bayesian error prediction by carrying out the slow version every <code>n</code> steps, by setting
One can still have Bayesian error prediction by carrying out the slow version every <code>n</code> steps, by setting
{{TAG|ML_IERR}}=n. But by default this is omitted.
{{TAG|ML_IERR}}=n. But by default this is omitted.
Since the calculation time of the fast version is of the same order of magnitude as the timing for the output of the molecular-dynamics results, we advise to decrease the output frequency for molecular dynamics. This is controlled by the tag {{TAG|ML_OUTBLOCK}}. By default it writes out all molecular-dynamics results at every molecular-dynamics step. Additionally the calculation and output of the pair-correlation function can be suppressed by setting {{TAG|ML_OUTPUT_MODE}}=0.


== Related tags and articles ==
== Related tags and articles ==

Revision as of 09:57, 21 November 2022

ML_LFAST = [logical]
Default: ML_LFAST = .FALSE. 

Description: This tag switches on the very fast execution mode for ML_ISTART=2.


Until now this tag is only available in the development version of VASP.

This tag significantly speeds up the calculation for ML_ISTART=2, but without the Bayesian error prediction. One can still have Bayesian error prediction by carrying out the slow version every n steps, by setting ML_IERR=n. But by default this is omitted.

Since the calculation time of the fast version is of the same order of magnitude as the timing for the output of the molecular-dynamics results, we advise to decrease the output frequency for molecular dynamics. This is controlled by the tag ML_OUTBLOCK. By default it writes out all molecular-dynamics results at every molecular-dynamics step. Additionally the calculation and output of the pair-correlation function can be suppressed by setting ML_OUTPUT_MODE=0.

Related tags and articles

ML_LMLFF, ML_ISTART, ML_IERR, ML_OUTBLOCK, ML_OUTPUT_MODE